灰色-隐马尔科夫下的信任路径筛选及聚合算法  

Path Selection and Aggregation Based on Grey & Hidden Markov Model

作　　者：李沛杰[1] 张兴明[1] 沈剑良[1] 

机构地区：[1]国家数字交换系统工程技术研究中心,郑州450002

出　　处：《小型微型计算机系统》2015年第5期964-970,共7页Journal of Chinese Computer Systems

基　　金：国家"八六三"高技术研究发展计划项目(2009AA012201)资助;上海市科委重大科技攻关项目(08dz501600)资助;河南省自然科学基金项目(122300413201)资助

摘　　要：针对现有信任模型在推荐路径的预测及选取方面存在的问题,提出一种基于灰色-隐马尔科夫模型(grey&hidden M arkov model,G&HM M)的信任路径筛选及聚合算法.首先对推荐系统的所有推荐节点进行灰色聚类分析,过滤恶意推荐路径;引入推荐接受程度衡量推荐节点的推荐可信度,然后基于灰色马尔科夫模型预测灰色推荐路径的推荐接受程度及其所在灰类状态;最后针对评价实体的观测序列构建隐马尔科夫模型,评估预测的最优推荐路径.仿真实验表明,与已有模型相比,该模型能够有效分离恶意推荐实体,提高推荐系统的有效性.Current trust models had some disadvantages in the prediction and selection of recommended path. A new path selection and aggregation strategy based on Grey ＆ Hidden Markov Model （ G＆HMM ） was proposed in this paper. In order to filter the malicious path, the model analyzed all the entities in the system using the grey clusters theory, and predicted the acceptation and grey state which the path had by the forecast of Grey Markov Model. The Hidden Markov Model was given to evaluate the optimal path finally. The experiment has shown that the proposed model can filter the malicious recommend entities and promote the effectiveness of the recommendation system.

关 键 词：推荐信任 推荐路径 灰色聚类 灰色马尔科夫 隐马尔科夫 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]